The Future of R&E Networks and Cyber-infrastructure

1. The Future of R&E Networks and Cyber-infrastructure Bill St. Arnaud Bill.st.arnaud@gmail.com Unless otherwise noted all material in this slide deck may be reproduced, modified or distributed without prior permission of the author

2. Looking Back • Internet (aka cyber-infrastructure) started with R&E networks world wide • Most popular Internet applications like web, Google, Facebook, etc started at universities connected to high speed R&E networks • Enabled by the permission free environment • R&E networks pioneered much of the global Internet infrastructure such as Internet exchange points, caching networks, etc • R&E networks pioneered the concept of customer owned dark fiber • Led to significantly lower costs for broadband and made The Netherlands world leader in consumer broadband

3. Impact of R&E networks & Cyber-infrastructure • OECD: “The Internet is the most powerful infrastructure we’ve seen yet for the creation, exchange, and implementation of new ideas. By facilitating access to information and collaboration at previously unimaginable scale, one small idea shared online can quickly grow into a world-changing one. As a result, the Internet has become a wellspring of innovation and renewal across the economy…” • R&E networks have enabled a trillion dollar industry, larger than transportation, construction and utilities combined • 13% of economic output and 6% of employment • No other R&E infrastructure or program has such a direct and measurable impact on society and the economy

4. NRENs will play an even more important role looking forward • Technologies developed at R&E networks today may have bigger impact on society then Internet had in the past • Big Data and Clouds • Global Collaboration • Software defined networks • Next generation Internet of integrated optical and wireless networks • Green networking

5. Changing R&Ecyber-infrastructure environment Global Virtual Research Communities Increasing potential of commercial ICT-service providers i.e. Clouds, storage, apps Education: any time, any place, any device Citizen Science and M2M communications and sensors Funding models changing from direct allocations to user pays networking computation can be purchased from commercial suppliers or R&E cyber-infrastructure Shrinking or disappearing IT department especially in small colleges

6. Issues and Challenges • Campus bottleneck for big data flows • Federated trust and Identity management • Campus remains main arbiter for identity management • The “cableization” of campus networks • TV and content traffic much greater than research traffic • Netflix can be 60% of campus traffic • Most traffic on-off rather than within campus • Biggest optical networks are run by clouds and content distribution networks • 98% of Internet content cachable • Google, Akamai, Facebook • How to interconnect to these facilities?

7. The Data Deluge 2004: 36 TB 2012: 2,300 TB Genomic sequencing output x2 every 9 month Climate model intercomparison project (CMIP) of the IPCC MACHO et al.: 1 TB Palomar: 3 TB 2MASS: 10 TB GALEX: 30 TB Sloan: 40 TB Pan-STARRS: 40,000 TB 1330molec. bio databases Nucleic Acids Research (96 in Jan 2001) Source: Ian Foster, UoChicago

8. Growth in sensor networks and Citizen Science Glacier Tracking Real Time Health Monitoring Smart Trash

9. Research Computing Pyramid Compute, compute, compute 102 Petascale/Exascale/… National HPC infrastructure Capable Users Cloud computing University HPC infrastructure Closet clusters 109 Mobile/Desktop computing Big Data Data, data, data Source: Dan Reed, PCAST

10. Small science is struggling More data, more complex data Ad-hoc solutions Inadequate software, hardware Data plan mandates Source: Ian Foster, UoChicago

11. The real cost of campus computing • Land - 2% • Core and shell costs – 9% • Architectural – 7% • Mechanical/Electrical – 82% • 16% increase/year since 2004 Belady, C., “In the Data Center, Power and Cooling Costs More than IT Equipment it Supports”, Electronics Cooling Magazine (February 2007) Source: Christian Belady

12. The ICT energy consumption inhigher- ed • Campus computing 20-40% electrical energy consumption on most campuses • Studies in UK and The Netherlands • http://goo.gl/k9Kib • Campus data center alone represents 8-20% of electrical consumption • http://www.iisd.org/publications/pub.aspx?pno=1341 • IISD study demonstrated that moving Canadian research to cloud would pay for itself in energy savings and CO2 reduction • http://www.iisd.org/publications/pub.aspx?pno=1341 • Internet and ICT can reduce CO2 emissions by 15-20% • No other tool has this potential

13. We need a new nuclear plant per day to keep stabilization at 2C Source: Roger Pielke Jr

14. How fast can decarbonization occur? • The honest answer no one knows • Historical rates of 1-2% have occurred in developing countries – mostly by outsourcing manufacturing to China • For short periods some countries have achieved rates > 2% e.g. France nuclear program • US need to achieve 17% reduction while maintaining modest economic growth which requires rates of decarbonization of >5% per year • Need to shut down over 20 coal plants per year to meet this target

15. NREN Brokered Cloud for IT departments and Researchers • Internet 2 Net + • Provisioning of multi vendor cloud services leveraging the Internet2 Network and InCommon Federated Authentication • Interoperable marketplace for services where individual institutions might procure services from a wide range of cloud services providers. • HEFCE and JISC to Deliver Cloud-Based Services for UK Research • Besides providing brokered cloud services they are also providing cloud “solutions” for IT departments and researchers • http://www.hpcinthecloud.com/hpccloud/2011-06-27/hefce_and_jisc_to_deliver_cloud-based_services_for_uk_research.html?utm_medium=twitter&utm_source=twitterfeed • SURFnet: Community Cloud Models and the Role of the R&E network as a broker for cloud services • http://www.slideshare.net/haroldteunissen/community-clouds-shared-infrastructure-as-a-service

16. Impact of NREN wireless networks • The phone is a also a sensor platform • Processing is done in real time in the cloud • Allowing processing that can’t be done on the device • Big data analysis • New campus or hot spot centric architectures integrating LTE and Wifi • See SURFnet pilot http://www.surfnet.nl/en/nieuws/Pages/Backgroundarticle.aspx

17. Three types of networks on campus • “General purpose“ IP network largely used by students and faculty for e-mail, web browsing, download videos, etc • IP router based • Often wireless WiFi • Consumes most of campus IT resources – security, routing, support, etc • “eScience” network used for large data transfers, high performance computing, big instruments, etc • Currently eScience underutilized because of choke points on campus network that frustrates many science users • Most need dedicated fiber or VPN on campus to DMZ with dedicated lightpaths to major eInfrastructure facilities • “Back office” university management computing/networking • Financial systems, Oracle, PeopleSoft, etc • Usually requires secure networks and/or VPNs • Many systems moving to cloud based services

18. Scientists are not network engineers • Studies by Esnet and Internet 2 indicate a large degree of frustration by scientists with network • Most scientists cannot do major file transfers or other data intensive work because of limitations of network on campus • Also tuning of servers etc • In a few cases, such as High Energy Physics, scientists have technical resources to get around campus network problems • To date most lightpath connections for scientists are mostly physics and astronomy • Most campus IT staff don’t have technical resources or inclination to provide special services for small number of eScience projects • Most campus IP networks not well suited for needs of eScience such as large file transfer

19. Disappearing Campus IT • Traditionally all networked applications and services existed on campus • e-mail, web, HPC, authentication, etc • Now most major applications exist off campus • Facebook, Twitter, Cloud apps, cloud computing • Grid computing, eScience, CERN, eVLBI, etc • Smart phones with 3G also bypass campus network • Collaborative tools and applications are being built outside of campus e.g. SURFconext, SURFmedia • Globalization of universities, education and research will continue this trend of moving services off campus • Universities with multiple campuses • Greatest volume of traffic is to and from outside of campus and not within • Universities are under increasing pressure to reduce costs • IT represents 20%-40% energy consumption on campus

20. You can run a company from a coffee shop

21. Towards “research IT as a service” Source: Ian Foster, UoChicago

22. Extending science and education to the community • Community anchor Internet Exchange Points help clear the bottleneck of content peering • Co-hosting of CDN caching boxes • Managed by NREN • Examples include KAREN (New Zealand), BCnet and UNINETT (Norway) • Minimize tromboning of R&E traffic to homes and schools • Can support extension of Eduroam to community WiFi spots and/or community last mile networks – see SunetTheCloudannouncement • Allows for M2M traffic and anywhere, anytime traffic to propagate through the community

23. Other national networks Regional National or Pan-National Network NREN A NREN C NREN B NREN D University Today’s hierarchical network

24. Regional Future Federated network Commercial Cloud/CDN Optical Network Other national networks National or Pan-National Network NREN A NREN C NREN B NREN D University

25. Campus view Campus DWDM University Global Physics Network 4G/Wifi Physics Department Commercial Clouds NREN No campus Network Health Network Firewall Global eVLBI Network University Hospital Radio Telescope Note: No campus network. Wifi Managed by NREN or commercial third party with integrated 4G/Wifi and Eduroam

26. Let’s Keep The Conversation Going E-mail list Bill.St.Arnaud@gmail.com Blogspot Bill St. Arnaud http://green-broadband.blogspot.com Twitter http://twitter.com/BillStArnaud